Image-based object classification method, system and electronic equipment

Abstract

The present invention provides an image-based object classification method. By obtaining a neural network with a rough classification model, the method adjusts various fine classification models to the corresponding coarse classification model, and saves the rough classification and fine classification of objects to be classified respectively. The feature vector is used as the retrieval bottom library, input the first image with the object to be classified, obtain n second images whose similarity with the first image is greater than the first threshold, and filter in the n second images to further obtain m A third image whose similarity with the first image is greater than the second threshold is subjected to ranging screening in the feature vectors of the m third images to obtain the classification result of the object to be classified, which improves the classification efficiency of the model, so that when When the neural network needs to identify a new type, it only needs to store the feature vector of the new type to perform identification and classification, avoiding the retraining of the neural network and reducing the dependence on a large number of learning samples.

Description

【Technical field】 [0001] The invention relates to the field of image recognition, in particular to an image-based object classification method, system and electronic equipment. 【Background technique】 [0002] In the field of image recognition, neural networks are usually used to identify and classify objects in images, which mark images of fixed types of objects and input the marked images to the model for training to obtain recognition and classification functions. neural network model. [0003] However, in the existing neural network with image recognition and classification functions, when it is necessary to recognize and classify new types of objects, it is necessary to prepare a large number of labeled training samples to retrain the neural network, which relies heavily on a large number of learning samples. , the classification efficiency is low. 【Content of invention】 [0004] In order to overcome the problem of low classification efficiency in the current existin...

AI Technical Summary

Problems solved by technology

[0004] In order to overcome the problem of low classification efficiency in the current existing image-based object classification methods, the present invention provides an image-based object classification method

Images